Implement and TestTable 2 Downloads

Table 2 Downloads

The specification's tables as machine-readable files, so you do not need to retype rules. They are generated from the specification document by script, not typed by hand, and cell text is exactly what the document contains.

The specification text governs. These files are a convenience copy of specification v3.4.0. They are regenerated for each version. If a file and the specification differ, the specification is correct.

TableRowsCSVJSON
Table 2: Category 1 rules30CSVJSON
Table 2-H: Household tier14CSVJSON
Table 2-I: Individual within household3CSVJSON
Category 2 combined probabilities (III.C.4)10CSVJSON
Administrative restrictions (III.C.6)2CSVJSON
Table 3: Conservative per-field u-probabilities21CSVJSON
Table 4: Sample administrative restrictions6CSVJSON

Reading the files

  • Values are strings exactly as printed, for example 3e-13, ≈0, or 2.0e-6 (exact) / 3.0e-6 (fuzzy). Parse them yourself. Empty cells are empty strings.
  • In Table 2, an asterisk after a field name marks a field that may be fuzzy. In the column "P(collision), fuzzy", an empty value means the row has no fuzzy variant.
  • The specification does not assign IDs to the Category 2 pairings in the III.C.4 table. Its rows appear in the specification's order. See the Spec-to-Code Map for how a reference implementation labels them.
  • Each JSON file records the specification version and the revision of the source document it was generated from.

Regenerating

tools/export_tables.py in the docs repository produces these files from the source document's Google Docs API JSON export. It is run by the maintainers for each new specification version; the repository's CLAUDE.md describes the inputs.

Was this page helpful?